google-deepmind / google-deepmind/gemma

[Feature Request] Improved Multilingual Support for Gemma 4

Open
#460 1 comment 1 reaction 0 assignees View on GitHub
Dominant language
Python
Stars
5.7k
Forks
1k
Avg merge
10h 33m
Merged PRs (30d)
2

Description

Hello Team,

While testing the current **Gemma 3n E2B, E4B** on-device models, I observed significant **performance degradation** when switching from English to **non-English languages**: Spanish and Russian. Specifically, the models just unable to maintain a conversation without making obvious mistakes/glitches.

It would be great to have more **production-ready** mobile models in the future that cover major global languages, not just the default ones, such as English for Gemma models or Chinese for Qwen family.

Current formal support of 140 languages in fact is very poor in Gemma 3n E2B/E4B.

High-quality coverage, at least for the main international languages, is essential for real mobile adoption in non-English markets.

Thanks and waiting for Gemma 4!

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.